ssprompt Download - ssprompt Source code download

ssprompt

AI Source Code

v0.0.5

Download

? ssprompt

Simplified Chinese | English

Change the world, even a little bit.

Since OpenAI launched a new AI revolution, many players at home and abroad have entered the market to receive the baptism of the times, and LLM-related technologies have exploded for a while, and prompt engineering (prompt engineering) is among them.
It's like a "language game" designed for Big Language Model (LLM). Through this "game", we can guide LLM more effectively to deal with problems. Before the arrival of real general intelligence, based on the current LLM paradigm, we must give full play to the advantages of LLM, Prompt design is becoming more and more complicated, and further the code of Prompt, modularization will become more and more obvious. At the same time, writing Prompt will become the basic skill of people in the AI era.
Based on this, I have conceived and created ssprompt, hoping that everyone can use Prompt and enjoy the dividends of the AI era.

Quick Install

System dependency

Ssprompt requires Python 3.10+

Supports multi-platform use to meet the majority of Prompt Engineers

pip installation

pip install ssprompt

Linux, macOS, Windows (WSL)

curl -sSL https://raw.githubusercontent.com/ptonlix/ssprompt/main/install.py | python3 -

Windows (Powershell)

(Invoke-WebRequest -Uri https://raw.githubusercontent.com/ptonlix/ssprompt/main/install.py -UseBasicParsing).Content | py -

? What it can do?

ssprompt is a Prompt distribution management tool that defines a set of Prompt distribution rules to support creating Prompt projects and pulling project files corresponding to Prompt on Prompt Hub to local projects.

Note: The following is the content introduction of the first version of ssprompt in 0.5.0. The project is still being improved. If there are any omissions or shortcomings, please forgive me for understanding. Thank you~

Command introduction

ssprompt new -> Create a new Prompt project, specify the project directory, Prompt type and other parameters
ssprompt init -> Based on the current directory, boot to create a Prompt project
ssprompt add -> Add a different Prompt type and related dependencies to the metafile and generate the corresponding Prompt project directory (project configuration file)
ssprompt show -> Show the basic information (metafile) of the local Prompt project or pull the information of the corresponding project on PromptHub
ssprompt pull -> Pull remote projects into local projects. The related projects can refer to Prompt files or codes
ssprompt list -> Display the commands supported by the current version of ssprompt
ssprompt about -> Show the introduction and version information of ssprompt
ssprompt version -> Display the version information of ssprompt

For more details of command parameters, please use ssprompt [command] -h to learn more about it

Introduction to Metafile

ssprompt constrains the management of Prompt distribution rules and content by defining the Meta file of the propt project
ssprompt About propt Define four types of Prompt

Text
Json
Yaml
Python

You can generate the corresponding Prompt as needed and upload it to PromptHub for distribution.
metafile named after the Prompt project name, such as propt_project.yaml, is the key to ssprompt managing Prompt distribution

Note: The above types are combined with reference to langchain and haystack.

 # Prompt工程基础信息
meta :
  name : open #工程名称
  author :
    - ptonlix <[email protected]>
  description : " "
  license : MIT # Prompt工程遵循的协议
  llm : # Prompt支持的LLM模型
    - gpt-3.5-turbo
  readme_format : md # Readme文件格式
  tag : # Prompt工程相关类型领域，如question-generation common为公共领域
    - common
  version : 0.1.0 #版本号

# Text类型的Prompt
text_prompt :
  dirname : text #目录名称, 默认为text

# Json类型的Prompt
json_prompt :
  dirname : json #目录名称，默认为json
  list : #支持多个json类型子工程
    - dependencies :
        langchain : 0.0.266 # json解析依赖的三方库版本号，如langchain等
      name : example #子工程名，对应生成工程目录名

# Yaml类型的prompt
yaml_prompt :
  dirname : yaml #目录名称，默认为yaml
  list : #支持多个yaml类型子工程
    - dependencies :
        langchain : 0.0.266 # yaml解析依赖的三方库版本号，如langchain等
      name : example #子工程名，对应生成工程目录名

# Python类型的Prompt
python_prompt : #目录名称，默认为yaml
  dirname : python #目录名称，默认为yaml
  list : #支持多个yaml类型子工程
    - dependencies :
        langchain : 0.0.266 # Python库引用的三方库版本号，如langchain等
      name : example #子工程名，对应生成工程目录名

Version dependency rules

The current version supports three version dependency rules

Caret requirements
- ^1.2.3
Tilde requirements
- ~1.2.3
Wildcard requirements
- 1.2.*
Laster
- Support the latest version

? PromptHub

Currently, the Prompt project generated by ssprompt relies on Git management. The Prompt project is uploaded to the Git repository through Git so that ssprompt can pull references

Current default PromptHub is hosted in GitHub ptonlix/PromptHub

At present, the Prompt project of PromptHub is still under construction?

More Prompt projects will be collected and built in the future and published to our PromptHub ☁️

? Everyone is welcome to upload their own Prompt to PromptHub to build an open source Prompt ecosystem

GitHub Token

Due to the current GitHub API request access restrictions, authentication access is not used, and it will limit access to only 60 times per hour, resulting in 403 restricted requests when frequently pulling projects with ssprompt

It is recommended to set GitHub Personal access tokens to environment variables when using ssprompt

 export GITHUB_ACCESS_KEY= ` Your GitHub Token `

Roadmap

? Project exchange and discussion

? Scan the code to contact the author. If you are also interested in this project, please join the ssprompt project group to participate in the discussion and exchange.